Adding evaluations for the LLM AI implementation of ChatState

[mahomedalid]

Overview

This PR enhances the eShop reference sample by incorporating evaluations for the implementation of ChatState utilizing OpenAI's Large Language Model (LLM) AI. The goal is to assess the functionality of the ChatState feature, particularly focusing in coherence, groundedness and relevance.

Key Additions:

1. Evaluations with xUnit Runner: Introduces evaluations for ChatState, leveraging xUnit as the test runner.
2. Integration with Azure OpenAI or local models: Adds configuration to run the evaluations using local models or Azure OpenAI.
3. GitHub Actions Workflow: Implements a GitHub Actions workflow to automate the testing process.

Notes:

• Fix an issue on ChatState that when a kernel is reused and ChatState constructor is called multiple times (for example on multiple initializations) it will throw an error when trying to add multiple times the same plugins/functions.

[luisquintanilla]

@mahomedalid looks great.

Just had one question.

[luisquintanilla]

@mahomedalid what is this for?

[luisquintanilla]